Payment & identity APIs are about to get a new user — the agent.
Coding agents already write most of the integration glue that ships software. The next step is autonomy over money: an agent reads an API's documentation, wires up the integration, and transacts — verifying identity, moving funds, settling — on a person's behalf, with no human in the loop.
When the user is an agent, the API an agent can read, trust, and operate wins — not the one with the best landing page. The Agent Usability Index ranks APIs on exactly that: how ready each one is for an agent to pick it up and use it.
Three signals, one ranking.
Agent rating
An agent reads the vendor's live documentation — API surface, llms.txt, SDK references — and we score how completely and unambiguously it can integrate and transact. It's the headline number on every row.
Posts
Each API gets a hosted, agent-readable blog. Posts is the depth of that corpus — structured content an agent can retrieve to learn the API, written and indexed for machine consumption, not human skimming.
Views
How much that agent-facing content actually gets read. As agents browse and cite documentation to finish a task, the docs built for them surface and get pulled. Views is our proxy for agent attention.
Each API is graded on how an agent experiences its docs.
Endpoints, parameters, and auth documented unambiguously enough to integrate without guessing.
An llms.txt (or equivalent) that lets an agent map the entire surface in a single read.
Idempotency, retries, and failure modes documented — not just the happy path.
Typed SDKs and copy-runnable code an agent can lift directly into an integration.
First-class support for how agents connect — e.g. MCP — not only human-facing REST.
Each dimension is scored and rolled into a single 0–100 agent rating, then mapped to the letter grade shown on every leaderboard row.
x402 integrations — measured throughput, not just readiness.
x402 — the HTTP 402 protocol agents use to pay — is coming to the index. As agents settle payments through each API over x402, we'll track the real dollar volume flowing through it by agents. That turns agent-readiness from a documentation proxy into measured throughput: which rails agents actually move money on.
